BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions,research libraries, and research funders in the common goal of maximizing access to critical research.

Grassland Carbon Sequestration Ability in China: A New Perspective fromTerrestrial Aridity ZonesAuthor(s): Yizhao Chen, Shaojie Mu, Zhengguo Sun, Chengcheng Gang, Jianlong Li, José Padarian,Pavel Groisman, Jingming Chen, Siwei LiSource: Rangeland Ecology & Management, 69(1):84-94.Published By: Society for Range ManagementURL: http://www.bioone.org/doi/full/10.1016/j.rama.2015.09.003

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological,and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and bookspublished by nonprofit societies, associations, museums, institutions, and presses.

Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance ofBioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use.

Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercialinquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.

Grassland Carbon Sequestration Ability in China: A New Perspective fromTerrestrial Aridity Zones☆

Yizhao Chen a, Shaojie Mu b, Zhengguo Sun c, Chengcheng Gang a, Jianlong Li d,⁎, José Padarian e,Pavel Groisman f, Jingming Chen g, Siwei Li h

a Associate Researcher, School of Life Science, Nanjing University, the People’s Republic of (PR) Chinab Researcher, Nanjing Institute of Environmental Sciences, Ministry of Protection, Nanjing, PR Chinac Associate Professor, College of Animal Science and Technology, Nanjing Agriculture University, Nanjing, PR Chinad Professor, School of Life Science, Nanjing University, the People’s Republic of (PR) Chinae Lecturer, Faculty of Agriculture and Environment, The University of Sydney, Sydney, Australiaf Professor, National Oceanic and Atmospheric Administration National Climatic Data Center, Asheville, North Carolina, NC 28801, USAg Professor, Department of Geography, University of Toronto, Toronto, ON, Canadah Postgraduate, School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, PR China

a b s t r a c ta r t i c l e i n f o

Article history:Received 10 January 2015Received in revised form 14 September 2015Accepted 28 September 2015

Keywords:aridity indexChinese grasslandterrestrial biogeochemical modelingterrestrial carbon cyclewater budget condition

Current climate change (e.g., temperature and precipitation variations) profoundly influences terrestrial vegeta-tion growth and production, ecosystem respiration, and nutrient circulation. Grasslands are sensitive to climatechange, and the carbon sequestration ability is closely related to water availability. However, how the terrestrialwater budget influences regional carbon sequestration by the grassland ecosystem is still unclear. In this study,we modified a terrestrial biogeochemical model to investigate net ecosystem productivity (NEP) of Chinesegrasslands under different aridity index (AI) levels from 1982 to 2008. The results showed that Chinesegrasslands acted as a carbon sink of 33.7 TgC. yr−1, with a clear decrease in the spatial distribution from thehumid end (near-forest) to the arid end (near-desert). During these 27 years, gross primary productivity(GPP) and net primary productivity (NPP) significantly increased with regional warming over the entire rangeof the AI, but no significant tendency was found for NEP. Meanwhile, only NPP in the arid zone (AR) and thesemiarid zone (SAR) were significantly correlated with mean annual precipitation (MAP), and no significantcorrelation was found between heterotrophic respiration (Rh) and MAP; NPP and Rh were both positively correlatedwith mean annual temperature (MAT) in all AI zones except for NPP in AR; no significant correlation between NEPandMAP orMATwas found. These results revealed that the grasslandswith different AI levels keep different responsepatterns to temperature and precipitation variations. On the basis of these results, we predicted that the gap of carbonsequestration ability between humid and arid grassland will expand. The total carbon sink in Chinese grasslands willcontinue to fluctuate, but there is a danger that it might shrink in the future because of a combination of climaticand human factors, although CO2 fertilization and N deposition might partly mitigate this reduction.

© 2016 Society for Range Management. Published by Elsevier Inc. All rights reserved.

Introduction

According to the latest Intergovernmental Panel on Climate Change re-port (2013), global surface temperature increased by 0.85°C from 1880 to2012 and will continue to rise by approximately 0.3−4.8°C in 2081–2100

compared with the temperature for 1986–2005. Given this situation, theglobal terrestrial carbon and hydrological cycles are experiencingand will continue to experience significant changes (Calzadilla et al.2013; Gosling and Arnell 2013). Furthermore, because of the tight interac-tions between these two cycles, theywill strongly affect each other aswell(Falkowski et al. 2000; Porporato et al. 2003).

Grasslands cover approximately 44% of the total terrestrial area inChina and comprise 13% of the world’s total grasslands (Ni 2002,2004a; Scurlock and Hall 1998). Grasslands are among the mostimportant terrestrial carbon pools in China, storing approximately3.06 Pg of vegetation carbon and 41.03 Pg of soil carbon (Ni 2002).The Chinese grasslands ecosystem is greatly affected by human activityvia land use and cover change, direct graze, and mowing, so it is also

Rangeland Ecology & Management 69 (2016) 84–94

☆ Research was funded by APN Global Change Fund Project (ARCP2013-16NMY-Li),the National Natural Science Foundation of China (41271361); The Key Project ofChinese National Programs for Fundamental Research and Development (973 Program,2010CB950702); the National High Technology Project (863 Plan, No. 2007AA10Z231);and the Public Sector Linkage Program supported by Australian Agency for InternationalDevelopment (64828).⁎ Correspondence: Jianlong Li, School of Life Science, Nanjing University, Hankou Road

22, Nanjing 210093, PR China.E-mail address: Jianlongli@gmail.com (J. Li).

http://dx.doi.org/10.1016/j.rama.2015.09.0031550-7424/© 2016 Society for Range Management. Published by Elsevier Inc. All rights reserved.

Contents lists available at ScienceDirect

Rangeland Ecology & Management

j ourna l homepage: ht tp: / /www.e lsev ie r .com/ locate/ rama

closely related to national agriculture and the animal husbandry system(Cao et al. 2011; Duan et al. 2011; Wang et al. 2011).

Grassland ecosystems are sensitive and vulnerable to global change(Fang et al. 2003; Ni 2003, 2004b; Piao et al. 2006c; Scurlock and Hall1998). The impacts of CO2 concentration, temperature, and precipita-tion variability can significantly affect the grassland carbon and waterbudget (Bai et al. 2010; Mitchell and Csillag 2001; Schimmelpfenniget al. 2014; Shen et al. 2011). Earlier regional results indicated that a sig-nificant upward trend ofmean annual temperature (MAT) coupledwithCO2 fertilization stimulated vegetation growth. Results from multiplestudies have shown that the net primary productivity (NPP) of Chinesegrassland increased by approximately 15−20% from 1980 to 2000,which is regarded as strong evidence for characterizing grasslands as acarbon sink (Fang et al. 2003; Ni 2000; Piao et al. 2003). During thepast several decades, Chinese grasslands acted as a carbon sink overall,and this trend is predicted to continue for the next 100 years (Fanget al. 2007; Ji et al. 2008; Piao et al. 2007). However,more recent studiesnoticed that an increase in soil respiration and water consumption dueto climate change could counteract the CO2 fertilization effect and evenreduce the current regional carbon sequestration in Chinese grasslands.Thus the carbon sequestration ability is likely to remain unchanged oreven shrink in the future, especially in those arid and semiarid areas(Lu et al. 2009; Piao et al. 2012b; Sui and Zhou 2013).

Previous studies that investigated the responses of the regional ter-restrial ecosystem to climate change were mostly based on categoriza-tion of the ecosystems by vegetation types, administrative division orgeographic zoning (Fang et al. 2007; Liu et al. 2013; Ni 2002; Yanget al. 2008b). In those studies, however, the terrestrial water availabilityhas not been explicitly identified, which could lead to confusion aboutcarbon and water interactions and their impact on regional terrestrialecosystems. For example, in the widely studied “arid/semiarid” area ofInner Mongolia, the water budget and the grassland types are diverse,ranging from the humidmeadow steppe in the northeast to an extremelyarid desert steppe in the southwest. Water availability varies greatlythroughout the region, so the response of these diverse ecosystems toclimate change should vary as well.

In this study, wemodified and validated a terrestrial biogeochemicalmodel, the boreal ecosystem productivity simulator (BEPS) (Chen et al.1999), to simulate vegetation primary production (gross primary pro-ductivity [GPP], NPP), soil respiration (Rh), and net carbon sequestrationability (net ecosystem productivity [NEP]) in Chinese grasslands from1982 to 2008. Themodel wasmodified to include the terrestrial carbon,hydrological, and nitrogen cycles and proved to work well in grasslandecosystems (Ju and Chen 2005; Ju et al. 2010; Matsushita and Tamura2002; Zhang et al., 2013a, 2013b). In this study, we further modifiedsome of the major model input parameters and algorithms to adjustthe model to better simulate the regional grassland ecosystem(e.g., the maximum carboxylation rate [Vc,max], autotrophic respiration,production allocation). We divided Chinese grasslands into four subre-gions according to the aridity index (AI), and we studied the terrestrialresponse to climate change in response to differentwater budget levels.The goals of this study are to (1) investigate the spatiotemporalvariation of the carbon sequestration ability in Chinese grasslandsfrom 1982 to 2008 using a modified terrestrial ecosystem model and2) quantify the response of Chinese grasslands to current climatevariations (i.e., temperature and precipitation variations) at differentterrestrial water budget levels (Fig. 1).

Methods

Model Description

In this study, we modified and applied a daily-step biogeochemicalmodel, the BEPS. Themodel was initially developed to calculate vegeta-tion GPP and NPP using the Farquhar leaf biochemical model and thecanopy double-leaf scheme (Chen et al. 1999). A terrestrial hydrological

model was incorporated to account for the soil water content (SWC)and evapotranspiration (Chen et al. 2007; Liu et al. 2003). In addition,a CENTURY-derived soil carbon and nitrogen model was introducedby Ju et al. (2006).

We modified the version of BEPS that includes all threemodules mentioned earlier. In this paper, we only describe the majormethods to calculate indexes of terrestrial carbon flux (i.e., NPP,heterotrophic respiration (Rh), and NEP), SWC simulation, and ourmodifications. A detailed introduction of the model can be found inthe Supplementary Material.

The NEP is the difference between NPP and Rh:

NEP ¼ NPP−Rh

where NPP is the difference between GPP and autotrophic respirationrate (Ra):

NPP ¼ GPP−Ra

For the GPP simulation, the leaf maximum carboxylation efficiency(Vc,max) is an important input parameter that controls photosyntheticassimilation (Warren et al. 2003). Instead of using a predefined value,we used a multienvironmental limits scheme to calculate the Vc,max

with the Farquhar model according to Zhang and Zhou (2012):

V c;max ¼ Vm;0 � f Tempð Þ � f SWð Þ � f CO2ð Þ � f SNð Þ

where Vm,0 is the maximum carboxylation rate under optimal environ-mental conditions and f(Temp), f(SW), f(CO2), and f(SN) are functions ofthe atmospheric temperature (Temp [°C]), surface SWC (SW [%]),atmospheric CO2 concentration (CO2 [Pa]), and soil nitrogen content(SN [mg/g]) that affects the Vc,max value. Ra is simulated as the sum ofgrowth respiration (Rg) and maintenance respiration (Rm). Rg iscalculated as a fixed percentage from GPP. For Rm simulation, in theprototype model, we replaced the original Bonan (1995) algorithmwith a plant component-specific scheme:

Rm;k ¼ Tk � rm;k � QT−Tre fð Þ

1010;r

Tk is the biomass of plant component k, and rm,k is the coefficient ofthe maintenance respiration. T and Tref are air temperature and refer-ence temperature for maintenance respiration, respectively. Q10,r is atemperature sensitivity factor for respiration calculated as a functionof temperature (Arora 2003):

Q10;r ¼ 3:22−0:046T

After NPP has been simulated, a production allocation model isadded according to Friedlingstein et al. (1999).

Rh is calculated by multiplying the carbon release rate (pk) from thetotal carbon amount of a soil carbon pool (Ck).

Rh ¼X8

k¼1

pkCk

For the Rh simulation, the major modifications of the soil modulefrom the previous version were as follows: 1) The woody litter poolwas excluded for grassland simulation and 2) the effect from soil mois-ture on decomposition (Fw) was calculated using the water-fill porespace (WFPS) according to the inTECmodel ver. 3.0 (Ju and Chen 2005):

Fw ¼ 5:44WFPS−5:03WFPS2−0:472

where WFPS is calculated as the ratio of SWC and saturated SWC.

85Y. Chen et al. / Rangeland Ecology & Management 69 (2016) 84–94

The SWC is calculated by total water inputs (rainfall and snowfall)minus the total outputs (e.g., evapotranspiration, canopy interception,interlayer interaction, runoff). To accomplish this, the soil is separatedinto three layers and the canopy is divided into sunlit and sunshadedparts, the same as for the photosynthesis calculation. The SWC is relatedto the carbon cycle by defining thewater stress effect on photosynthesisand soil respiration.

The revised grassland specific input parameters are shownin Table 1.

Data Preparation

Aridity Index (AI) Classification MapWe used the AI map from the Consultative Group for International

Agriculture Research (CGIAR) from 2009 (aridity index dataset:http://www.cgiar-csi.org/data/global-aridity-and-pet-database).

In this product, the AI was defined as the ratio of mean annual precipi-tation (MAP) to theMeanAnnual Potential Evapotranspiration (MAPE):

AI ¼ MAP=MAPE

MAP is obtained from the WorldClim Global Climate Data (Hijmanset al. 2005). The temporal scale is from 1950 to 2000. MAPEwas modeled using the Hargreaves and Allen (2003) algorithm usingthe mean temperature and extraterrestrial radiation (RA), which wascalculated by the method of Allen et al. (1998) method. The spatialresolution of the product is 0.008333 degrees. It was resampled into8-km resolution.

According to the original dataset, terrestrial ecosystems are classi-fied into five levels: hyper-arid (AI b 0.03, HAR), arid (0.03 b AI b 0.20,AR), semiarid (0.20 b AI b 0.50, SAR), dry and subhumid (0.50 b AI b0.65, DSH), and humid (AI N 0.65, HU). However, few hyper-arid

Fig. 1. Chinese grassland under different aridity levels.

Table 1Grassland specific input parameters for the improved model.

Parameter Abbreviation Original Value Value Description Source

Vmax,opt / 100 (mmol m−2 s−1) Optimal maximum rubisco-limited potential photosynthetic capacity (Li et al., 2011)SLA 0.015 (m2 g−1) 0.025 (m2 g−1) Specific leaf area (Schapendonk et al., 1998)LR co. 0.008 0.002 Leaf respiration (Matsushita and Tamura, 2002)

CoefficientSR co. 0.001 0.00005 Stem respiration (Matsushita and Tamura, 2002)

CoefficientRR co. 0.003 0.0002 Root respiration (Matsushita and Tamura, 2002)

Coefficient

86 Y. Chen et al. / Rangeland Ecology & Management 69 (2016) 84–94

Chinese grasslands exist, so we combined that category with AR(see Fig. 1).

Model Input Data

Land Cover MapThe MODIS land cover product (USGS-EROS, MODIS product,

MOD12Q1-2001, Available at: https://lpdaac.usgs.gov) was used dueto its good quality and wide usage. We used the version with IGBPglobal vegetation classification and 1-km resolution. Chinese grasslandmap was extracted and resampled to 8-km resolution to fit for themodel input.

LAIWe used a daily-step 8-km LAI product (CAS, Global LAI database,

Available at: http://www.globalmapping.org/globalLAI/) to drivethe model. The consistent 8-km daily LAI product was derived fromboth Moderate Resolution Imaging Spectroradiometer (MODIS) andAdvanced Very High Resolution Radiometer (AVHRR) products. TheMODIS LAI series was generated using GLOBCARBON LAI algorithm forthe period since 2000 (2000–2008 in this study). Thereafter, the rela-tionships between AVHRR SR and MODIS LAI were established pixelby pixel during the overlapped period (2000–2006). On the basis ofthese relationships, homogeneous LAI serieswere retrieved fromhistor-ical AVHRR observations for the period before 2000 (1982–1999 in thisstudy). More details can be found in Liu et al. (2012).

Daily Meteorological DataThe daily air temperature, precipitation, radiation, and specific hu-

midity are required to drive themodel.We used the GlobalMeteorolog-ical Forcing Dataset for Land Surface Modeling (Princeton University,Global Meteorological Forcing Dataset for Land SurfaceModeling, Avail-able at: http://rda.ucar.edu/datasets/ds314.0/). This database is basedon global observation datasets and NCEP/NCAR reanalysis, and toenhance its accuracy, it was recently updated with the results fromthe World Meteorological Organization (WMO) Solid PrecipitationMeasurement Inter-comparison and Global Precipitation ClimatologyProject (GPCP) daily product. Thereafter, the data were evaluated byusing the second Global Soil Wetness Project (GSWP-2) dataset and753 basic meteorological stations across China. The spatial resolutionis 8 km.

Atmospheric CO2 DataMonthly atmospheric CO2 data were collected from Mauna Loa

Observatory (MLO), Hawaii (20°N, 156°W) (MLO, Air CO2 database,Available at: http://cdiac.esd.ornl.gov/ftp/trends/co2/maunaloa.co2).

Soil Texture DataSoil texture data were collected from the Global Soil Dataset for use

in Earth System Models (GSDE, Available at: http://globalchange.bnu.edu.cn/research/soilw). The data are displayed in the volumetricpercentages of silt, clay, and sand, from which hydrological parametersincluding the wilting point, field capacity, soil porosity, and maximumwater-holding capacity can be estimated. The spatial resolution is 1 km.

All the input datasetswere extracted and resampled to 8 km to fit forthe model simulation.

Model Simulation and Data Analysis

Themodel uses a semianalytical scheme to calculate the steady stateof soil carbon storage (Ju et al. 2006; Xia et al. 2012). First it calculatesthe regional quasi-steady-state pool sizes by setting the terrestrial fluxequal to zero to ensure that the pools’ sizes are close to the assumedsteady state without running the model for hundreds or thousands ofyears. Next, the pools are further spun up by using mean historicalforcing data to meet the steady-state criterion. The run time depends

on the carbon residence time in the various pools. In this study, weused the daily climatic data during 1948–1953 and a correspondingCO2 concentration of 311 ppm to execute the initialization. Finally, weran the model with transient climatic forcing and a rising CO2 concen-tration to obtain the carbon storage level at the beginning of 1982. Dur-ing this period, LAI was estimated using the specific leaf area (SLA, theratio of leaf area to leaf biomass) and simulated leaf biomass. A param-eter of maximum LAI was used to constrain the LAI upper limit. Afterthat, themodelwas run againwith the historical LAI andmeteorologicalinputs from 1982 to 2008 with a daily step.

We defined a linear regression model y = a + bx to analyze thespatiotemporal trend of GPP, NPP, and NEP, where x represents thestudy year.

A one-way analysis of variance (ANOVA) was used to test the signif-icance of correlationships between ecosystem indexes (i.e., NPP, NEP,and Rh) and climatic factors (i.e., MAP and MAT) using SPSS 20.0 (IBMSPSS Inc., Chicago, IL, USA). Differences at p b 0.05 and p b 0.01 wereconsidered as significant and very significant, respectively.

Model Validation

Simulation results for both the original and modified models werecompared with EC measurements and field work results (Table 2).

The two Eddy Covariance (EC) sites are located at Xilinhot (IM) andTongyu (TY) (Chen et al. 2009; Yang et al. 2008a). Daily-step data for GPPandnet ecosystememission (NEE, equal to−NEP)with completemeteoro-logical information were used to evaluate the corresponding model results.

Field measurements of biomass were collected from major grass-lands in China, including grasslands in Xinjiang, Inner Mongolia andon the Qinhai-Tibet Plateau. Each of these sites represents a typical veg-etation pattern in a local area. The biomass datawere sampled from fourreplicates within each test site. The total biomass was defined as thesum of the herbaceous, litter, and root biomass. The peak-season livingaboveground biomass and litter was measured by the destructive sam-pling of 1.0m2. The recent litters were selected. Thereafter, the sampleswere dried at 65°C to constant weight and then weighted. Roots werecollected by excavating 1.0m2 squares to a depth of 0.5m. Rootmatterswere hand sorted from soil, washed, dried at 65°C, and weighted. Thedry matter values were converted to carbon assuming a proportionalfactor of 0.475 (Garbulsky and Paruelo 2004).

Information about the sites can be found in Table 2.

Results

Model Validation and Comparison

Themodifiedmodel had better performance than the originalmodelfor both comparisons with the EC sites and field studies (Figs. 2 and 3).For the EC sites, the average R2 values of GPP (GPPMOD) against GPPfrom EC measurements (GPPEC) were increased by 0.15 (0.77 and 0.83for the revised model compared with 0.64 and 0.66 for the originalmodel in IM and TY, respectively). The average R2 increase in the netecosystem emission (NEE, equal to −NEP) was 0.11 (0.84 and 0.78 forthe revised model compared with 0.73 and 0.68 for the original modelin IM and TY, respectively). The average RMSE values of GPP decreasedby 0.66 gC.m−2 (1.25 gC.m−2 and 1.4 gC.m−2 for the revised modelcompared with 1.99 gC.m−2 and 1.96 gC.m−2 for the original modelin IM and TY, respectively), and the value decreased by 0.21 gC.m−2

for NEE (0.96 gC.m−2 and 0.92 gC.m−2 for the revisedmodel comparedto 1.21 gC.m−2 and 1.1 gC.m−2 for the original model in IM and TY,respectively). The modified model also showed a better correlationwith the field NPP (R2 = 0.79 for the modified model against 0.76 forthe original model, P b 0.001, see Fig. 3).

On the basis of these statistics comparisons, we believe that ourmodifications could improve the model performance for Chinese grass-lands. Themodel results are credible and suitable for this regional study.

87Y. Chen et al. / Rangeland Ecology & Management 69 (2016) 84–94

Spatiotemporal Pattern of GPP, NPP, and NEP at Different AIs

Fig. 4 shows the spatial distribution of the mean annual value, slope(b), and correlation coefficient (r) of GPP, NPP, and NEP in China from1982 to 2008. GPP and NPP have similar spatial distributions and varia-tion trends for Chinese grasslands, indicating a relatively stable autotro-phic respiration level. Grassland production has a clear zonaldistribution from near-forest to near-desert. High-productivity grass-lands are mainly located in HU, DSH, and those SAR areas connectedto forests, such as themeadow steppe joined to Great Khingan and Sibe-ria Forest in the eastern and southeastern part of the Mongol Steppe,and the alpine meadow near the southern and eastern part of theQinghai-Tibet Plateau. The values for those areas are generally above200 and 150 gC.m−2 per year for GPP and NPP, respectively. Theseparts are also among the fastest-growing areas with a significant posi-tive trend (see Fig. 4b and c). Grasslands with lower productivity arethose areas near the Gobi and Taklimakan deserts, for which GPP andNPP are typically b 20 and 10 gC.m−2 per year, respectively. A largepart of these areas also experienced a significant production increase;however, the absolute value was strongly limited by the environmentalcondition and the intrinsic vegetation growth potential. The main re-ductions were concentrated around the northwestern part of theTibet-Qinghai Plateau and the northeastern corner of Inner Mongolia

extending to the central part, but the decreasing trend was mostlyinsignificant.

NEP calculated by themodel has a similar spatial distribution to NPP.Chinese grasslands mainly acted as a carbon sink during the 27 yearsunder consideration (see Fig. 4a). Strong sink areas are those withhigh productivity, andweak sink or source areas are thosewith lowpro-ductivity. The regional temporal trend showed a divergence with pro-duction (see Fig. 4c). A decreasing trend was found in the northernpart of Xinjiang Province, the northern and eastern part of InnerMongolia, and the southeastern part of the Qinghai-Tibet Plateau.However, only small parts of these areas showed significant changes.A significant increasing tendency was found in the SAR areas nearTaklimakan desert and in southern Inner Mongolia.

Temporal Variations of Meteorological Factors and TerrestrialCarbon Flux Indexes

During the research period, the entire region’sMAT increased by ap-proximately 1.3°C according to the fitted linear regression equation,showing significant trends in all four subregions (Fig. 5a). However,no significant MAP trend was found except a slight increase in the ARregion (Fig. 5b). Under such environmental conditions, Chinesegrasslands’ productivity increased in all AI zones. The overall

Table 2List of site information for model validation.

Site Data type Long. Lati. Climate⁎ Time extent Number of samples

IM FLUX 116°40'E 43°33'N BSk 2004 /TY FLUX 122°52'E 44°25'N Dwa 2008 /Northern Tibet Field 100°51’-101°17’E 36°57’-37°36’N ET 2004-2007 49Xinjiang Field 88°37'E-88°40'E 44°29'N-44°31’N Bwk 2012 52Inner Mongolia Field 111°6'3"E-118°20'24"E 42°19'12"N −46°9'36"N BSk, Dwb 2004-2008 54

⁎ Climate type of grassland is based on Koeppen-Geiger classification (http://koeppen-geiger.vu-wien.ac.at/). BSk: main climate— arid, precipitation— steppe and temperature— coldarid; Dwa:main climate— snow, precipitation—desert and temperature—hot arid; ET:main climate—polar and temperature—polar tundra; Bwk:main climate— arid, precipitation—desertand temperature— cold arid; Dwb: main climate— snow, precipitation— desert, temperature—warm summer.

Fig. 2. Comparisons ofmodeled GPP (GPPMOD, gC.m−2.d−1) and NEE (NEEMOD, gC.m−2.d−1) to daily EC sites’measurements in Tongyu (TY) and InnerMongolia (IM): (a) the results withthe original model (OR); (b) the results with revised model (RE). The unit of RMSE is gC.m−2.

88 Y. Chen et al. / Rangeland Ecology & Management 69 (2016) 84–94

increments were about 50.1 gC.m−2 and 36.1 gC.m−2 for GPP and NPP,respectively (Fig. 5c and d). However, during the period, NEP showed acontinuous fluctuation without any significant trend (Fig. 5e).

Impact of Changes in Meteorological Factors on NPP, Rh, and NEP

The responses of NPP and Rh to climate change in the different AIzones are shown in Table 3. Significant relationships exist betweenNPP andMAT in HU, DSH, and SAR zones, with relatively good explana-tion ability. Slope increases from SAR to HU from 12.33 to 29.18 wereevident. Rh also showed significantly positive correlation with MAT inall AI zones and slope increases from AR to HU from 6.50 to 26.91.

The linear correlations between NPP andMAP become progressivelyweaker from the arid end to the humid end. Significant correlationswere found for AR and SAR but not for DSH and HU. In contrast, wedid not find a significant correlation between Rh and MAP across theentire range of AI levels.

Moreover, no direct correlation was found between NEP and MATor MAP.

Discussions

Our modified model performed better for grasslands than the origi-nal BEPSmodel at both EC sites and for fieldmeasurement comparisons.From a regional perspective, our NPP result is generally consistent withthe results ofmajormodels butmuch less than the result from the CASAmodel (Potter et al. 1993) (Table 4). However, it seems that the originalCASAmodel overestimatedNPP because the value declined significantlyafter an improvement byXing et al. (2010) andwas close to our result aswell. For NEP, we found that our result was similar to the result from theGLPM and slightly higher than the result from the TEM. The main rea-sons for these differences could be the longer study period in the Suistudy (1951–2007), which covered 3 decades during which NEP waslower than in recent times. Furthermore, different model inputs couldalso influence the results. Our result is much higher than that derivedfrom the grassland resource inventory. We think that the use of differ-ent methods to calculate vegetation production is one of the main rea-sons for the discrepancy in the results. In the work of Fang et al.(2007), the aboveground biomasswasdirectly estimated via a statisticalmodel from the annual maximum NDVI and the grassland biomass Cdensity, and the C content was derived from the biomass using a factorof 0.45. Belowground biomass is estimated by a predefined root-to-shoot ratio for each grassland type. However, for BEPS and similar

terrestrial biogeochemical models, the total carbon input is estimatedby the Farquhar biochemical model or a light-use efficiency schemeand ecosystem respiration is estimated using a multipool scheme. Thismechanical difference in estimation methodology has produced largediscrepancies in regional studies. According to the study by Piao et al.(2012a), an estimation using the ensemble terrestrial ecosystemmodel (without considering biofuel consumption) is higher than the re-sult from the inventory method for East Asia (−0.413 ± 0.141 PgC.yr−1 vs. −0.293 ± 0.033 Pg C.yr−1). Moreover, the different landcover classifications could also account for the discrepancies. In Fang’sstudy, more barren regions were classified as grasslands, especiallyarid/semiarid areas located in Central and Northwestern Tibet (closeto the Taklimakan Desert). These areas are characterized as desertsteppe and have a very low carbon density and sequestration potential.

In this study, the AI classification distinguishes Chinese grasslandswith different aridity levels and reveals the different situations, trends,and responses to climate change. Induced by various biogeographic fea-tures of grasslands in different AI zones, the response pattern of thegrasslands to hydrothermal conditions varies. Different responsesfrom grassland toMAP could be explained by a change in themajor lim-itation for vegetation growth. For relatively humid areas, precipitation isalways sufficient for regular vegetation growth and transpiration, sotoomuchwater input can induce ineffectivewater use and a higher bio-geochemical requirement, such as a nutrient limitation (Austin andVitousek 1998). For arid/semiarid areas, vegetation production is seri-ously limited by water scarcity and induces vegetational limitationssuch as low LAI and high root-to-shoot ratio (Bai et al. 2010; Fan et al.2009). Thus vegetation production is much more sensitive to wateravailability in such areas. Our results for arid/semiarid areas are in ac-cordance with those of Yang et al. (2010) and Ni (2004a), who alsonoted a significant positive correlation between NPP and MAP. In con-trast, we do not find a direct correlation betweenMAP and Rh. One pos-sible explanation is that MAP hardly represents the true soil moistureconditions because moisture is consumed by other means such asevapotranspiration or runoff and less is conserved in the soil, especiallyin arid/semiarid areas. In addition, the intra-annual precipitation distri-bution could have a profound impact on this response. If more extremerainfall events occur, the actual period for which the soil water contentis suitable could be reduced in humid ecosystems, while it could beprolonged in arid ecosystems. This mechanism has been theoreticallyexplained and demonstrated by multiple studies and might contributeto the ambiguous relationship between MAP and Rh (Jentsch andBeierkuhnlein 2008; Knapp et al. 2008; Luo et al. 2008). For MAT,

Fig. 3. Comparison between modeled NPP (NPPMOD, gC.m−2) and observed NPP (NPPField, gC.m−2) from independent field measurements collected from Inner Mongolia, Qinghai-TibetPlateau and Xingjiang: (a) the result with the original model (OR); (b) the result with revised model (RE).

89Y. Chen et al. / Rangeland Ecology & Management 69 (2016) 84–94

significant positive production correlations were found in most areas,except for the AR zone. This suggests that a difference still exists be-tween the present temperature conditions and optimal temperatureconditions for grass growth. According to the results from Piao et al.(2006a) and Parmesan (2007), an overall increase in temperaturecould promote earlier greening and extend the growth season in tem-perate grasslands, especially in these areas that have a high growth po-tential (Knapp and Smith 2001). Additionally, higher temperaturescould also promote soil N mineralization and therefore benefit the nu-trition supply for vegetation (Melillo et al. 2002).

Chinese grasslands mainly consist of SAR (N50%) with the largestamount of carbon sequestration of approximately 16.4 TgC per yearand the HU zone with approximately 9.5 TgC per year. These twozones comprise the major contributors to the carbon sink in Chinesegrasslands. Under the current trend of climate warming, a concurrentincrease in vegetation production and soil respiration leads to a fluctu-ation in carbon sequestration. We did not find a direct relationship be-tween NEP and MAT, but significant a positive correlation betweenNPP/Rh and MAT was found, and the Rh was found to increase morequickly than NPP in the AR and SAR zones (70% of the entire area) butmore slowly in the DSH and HU zones (≈30% of the entire area) as

MAT increased. Then, we calculate the net increase of MAT-drivenNPP and Rh over the entire region by multiplying them by the areas ofeach AI (we assumed that no NPP variation occurred in AR with aMAT increase). The results showed that Chinese grasslands could lose1.29 TgC per year with each 1°C rise in MAT, whichmeans that regionalwarmingwill lead to larger differences in carbon sequestration betweenarid and humid grasslands and an overall reduction of the grassland car-bon sequestration ability. The result is in accordance with previousstudies by Piao et al. (2012b) and Lu et al. (2009), who also remarkedthatwithout the influence of other unforeseen factors, grassland carbonfixation ability will probably weaken due to climatic warming in the fu-ture, especially in arid/semiarid areas. A possible increase in precipita-tion during the growth season may slightly mitigate this trend bystimulating productivity in the AR and SAR zones. But the extra waterinputmay also indirectly stimulate themicrobial respiration by creatinga more suitable soil environment (Janssens and Pilegaard 2003; Penget al. 2009; Zheng et al. 2009).

The current atmospheric CO2 concentration is not at the saturationpoint for plant photosynthesis, so an increase could still cause increasedbiomass accumulation, which has been demonstrated in some regionalstudies (Piao et al. 2006b; Sitch et al. 2013). At the same time, the

Fig. 4. Spatial distribution of (a)mean annual value (MEAN), (b) annual change rate (SLOPE) and (c) correlation coefficient (R) of GPP, NPP and NEP during the period from 1982 to 2008.Positive (negative) values of SLOPE and R represent increase (decrease) of the corresponding index; p values represent significant levels.

90 Y. Chen et al. / Rangeland Ecology & Management 69 (2016) 84–94

Fig. 5. Temporal trends of (a) MAT(°C), (b) MAP (mm), (c) GPP (gC.m−2.yr−1), (d) NPP (gC.m−2.yr−1) and (e) NEP (gC.m−2.yr−1) in different AI zones during the period from 1982 to2008. Green lines representHUzone, yellow lines represent DSH zone, orange lines represent SAR zone,maroon lines represent AR zone andblack lines represent regionalmeans. The solidstraight lines represent that the significant linear trend (p b 0.05) exists during the period.

Table 3Linear correlationships between annual NPP/Rh and MAT/MAP.

MAP MAT

NPP Equation R2 p Equation R2 p

AR y = 0.1498x + 4.0693 0.4014 b0.001 NSAR y = 0.1067x + 59.78 0.1558 b0.05 y = 12.328x + 73.522 0.4251 b0.001DSH N y = 19.695x + 151.71 0.5139 b0.001HU N y = 29.182x + 223.81 0.5107 b0.001

Rh Equation R2 p Equation R2 pAR N y = 6.4972x + 19.678 0.5426 b0.001SAR N y = 13.473x + 53.486 0.6622 b0.001DSH N y = 18.742x + 126.56 0.6309 b0.001HU N y = 26.907x + 183.79 0.455 b0.001

“N” represents no significant relationship exists, i.e. p N 0.05.

91Y. Chen et al. / Rangeland Ecology & Management 69 (2016) 84–94

studies in both recent decades and future predictions agree that thegreatest part of the carbon sink attributed to a rise in CO2 is distributedin Southeast China (Sui et al. 2013; Wang et al. 2014). Hence this re-sponse patternmay contribute to larger differences in carbon sequestra-tion between arid and humid areas. Moreover, it should be noted thatCO2 fertilization may only play a minor role in the NEP variation of Chi-nese grasslands. For example, Mu et al. (2008) indicated that althoughCO2 fertilization could have a strong impact on the Chinese carbonsink, its impact on grasslands is the weakest and only accounts for0.3% of the total NPP. The results from the TEM showed that CO2 fertil-ization could only counteract 1.4% of a NEP decrease while the climaticfactor caused a NEP variation of 15.3% (Sui et al. 2013).

N deposition could enhance photosynthesis input by increasing theleaf nitrogen levels and decelerate soil carbon decomposition by raisingthe ratio of C and N in soil pools. According to a summary by Piao et al.(2012a), N deposition could lead to an increase in carbon sequestrationin China ranging from −0.125 Pg C to b −0.089 Pg C. Currently, moststudies on Chinese grasslands are focused on semiarid temperatesteppes (e.g., InnerMongolia), while little work has been done on grass-land areas on a national scale. The limitations come from a paucity offield research and the existence of large uncertainties in nitrogen-coupled model simulations (Liu et al. 2011).

By incorporating LAI to drive the model, we implicitly include theimpacts from some major disturbances in grasslands, such as grazing,mowing, and fire. On the basis of this calculation, vegetation responsesto disturbances are actually added into the production prediction andthe bias range could largely depend on how much productivity is con-sumed or lost to the disturbances. In the past 30 years, the number oflivestock in China has exploded. According to our previous research inInner Mongolia, the total number of livestock in 2010 more than dou-bled from that in 1986 (≈100 million head vs. 40 million head, respec-tively), which largely aggravated grazing and mowing pressure ongrasslands and led to continuous degradation (Mu et al. 2013). This sit-uation is currently common in the SAR and AR zones in China, such as inthe temperate steppe in the Mongol Plateau and the alpine meadowgrasslands in the northwestern part of the Qinghai-Tibet Plateau (Liet al. 2005; Li et al. 2006). According to the surveys of Liu et al.(2004), approximately 90% of Chinese natural grasslands is degradedto various extents. As one thing leads to another, this trend will finallyresult in an undesirable transformation of the land cover from grass-lands to sand land or desert (Pei et al. 2008). In addition, other distur-bances such as mining and fires on grasslands are predicted toincrease (Liu and Diamond 2005; Stephens et al. 2014; Wang et al.2013). Thus a direct loss in productivity major perturbations increasedfrom the 1980s to 2000s, and these trends will probably continuein the future due to the increasing demands of animal production(McMichael et al. 2007). Because the aggravation of regional degrada-tion has been noted in major Chinese pasture areas (especially duringthe period 1980–2000) (Zhang et al., 2013a), we think that regional dis-turbances negatively affect grassland ecosystems. Therefore the pure

NPP responses to climate change are actually greater than our currentcalculations indicate. This bias could also be present in former regionalsimulations and predictions based on diagnostic models because themodel inputs for canopy data are similar (e.g., LAI, NDVI, vegetation cov-erage) for representing photosynthetic capacity and vegetation growth.Therefore we think that stronger production responses to climatechange should be noted in future studies, and a comprehensivedescription of terrestrial biogeochemical cycle responses to distur-bances is urgently needed for large-scale ecosystem carbon budgetmodeling and calculations.

Land use and cover change (LUCC) is another unpredictable factor.The current LUCC is closely related to human activity and governmentpolicies. On the onehand, inappropriate human use and activities, in ad-dition to the increasing frequency of extreme climatic events(e.g., drought, snowstorms), have contributed to grassland degradationin arid/semiarid areas, as discussed in the previous paragraph. On theother hand, the Chinese government has instituted several large regionalprograms to address these problems since the late 1990s. One of themost well-known projects is the Grain to Green Project (GTGP), whichhas induced a land conversion from deserts and croplands to grasslandsor forests (Wang et al. 2007). Reports indicate that the grassland area in-creased by 77,993 km2 from 2000 to 2009, with 29,432.71 Gg C. yr−1 inInner Mongolia (Mu et al. 2013). A similar trend has been reportedfor Shanxi, Gansu, Qinghai, and Xinjiang provinces (Feng et al. 2005).Moreover, the current trend of urbanization has accelerated the expan-sion of urban areas to original natural ecosystems, and this process alsomoves a large amount of the rural population to the cities, thus contrib-uting to the recovery of grasslands and shrublands in rural areas (Pärtelet al. 2007; Peng et al. 2009). According to a summary by Liu et al. (2010),climate warming and the pursuit of economic profits are the majordrivers of grassland reduction, while large ecological projects, forexample, the GTGP, drive grassland expansion in China. Therefore wethink that a combination of these factors accounts for part of the largeuncertainty regarding carbon sequestration in Chinese grasslands.

Above all, on the basis of our predictions, the carbon sink ability ofChinese grasslands will continue to fluctuate in the future, but a dangerexists that this sink might be minimized because of multiply factors.

Implications

According to the results from our model, under current climatechange trends, grassland production increased in all AI zones from1982 to 2008. However, because soil respiration was simultaneouslyincreased, the grassland carbon sequestration ability fluctuated anddid not show a significant trend. Meanwhile, grassland productionwas only sensitive to MAP in arid/semiarid areas, while a higher MATled to production increases in a wider range of grasslands because ofthe large difference between present temperature conditions and theoptimum temperature conditions for grasslands. The warming trendled to an Rh increase as well. However, no direct correlation was

Table 4Result comparison between recent studies and this paper of Chinese grassland’s NPP (gC.m−2.yr−1) and NEP (gC.m−2.yr−1).

Study area Study period Method Results (gC.m−2.yr−1) Corresponding results in this paper⁎

(gC.m−2.yr−1)reference

Northern China 2000-2005 Improved CASA model NPP:153.26 143.4 (Xing et al., 2010)China 2000 CASA NPP:245 141.58 (Gao and Liu, 2008)

CEVSA NPP:208GLOPEM NPP:145GEOLUE NPP:178GEOPRO NPP:168

China 2007 BEPS NPP: 122.6 155.72 (Feng et al., 2007)Xilin River basin, Inner Mongolia 2002 GLPM NEP:1.91 2.23 (Zhang et al., 2009)Inner Mongolia 1951-2007 TEM NEP:11.25 15.97 (from 1982 to 2007) (Sui et al., 2013)China 1981-2000 Chinese Grassland resource inventory NEP:2.12 5.65 (from 1982 to 2000) (Fang et al., 2007)

⁎ the results are extracted from the simulated map in this study with the same spatiotemporal scale to the corresponding recent studies. The results are mean annual values of thecorresponding study period.

92 Y. Chen et al. / Rangeland Ecology & Management 69 (2016) 84–94

found between NEP and MAP or MAT. On the basis of these results, wepredict that a danger exists: Chinese grasslandsmay exhibit a lower car-bon sequestration ability in the future in response to multiple factors,including those arising from terrestrial biogeochemical cycles, distur-bances, and LUCC. The difference between the carbon sequestrationability between arid and humid zones will probably increase. In addi-tion, we also suggest that to accurately quantify regional carbon seques-tration ability and its responses to climate change, a comprehensivedisturbances model should be incorporated into current terrestrial eco-system modeling.

This paper could contribute to an in-depth understanding of carbonsequestration and response mechanisms in Chinese grasslands under abackground of global change and be a reference for policymakers to usein the development of further plans and projects to protect grasslandsand preserve their functions in terrestrial carbon circulation.

Acknowledgments

We are sincerely grateful to the editor and anonymous reviewers.We also greatly acknowledge Dr. Ju Weimin from Nanjing Universityto offer the original model, data, and instructions; Dr. Shi Zhengfrom Oklahoma University to suggest revisions for the paper; andDr. Fu Congbin and Dr. Meng Xiangxin from the Chinese Academy ofScience to provide EC data in Tongyu. Without their kind help, thiswork would not have been finished.

References

Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration—Guidelines forcomputing cropwater requirements-FAO Irrigation and drainage paper 56. 300. FAO,Rome, p. 6541.

Arora, V.K., 2003. Simulating energy and carbon fluxes over winter wheat using coupledland surface and terrestrial ecosystem models. Agric. For. Meteorol. 118, 21–47.

Austin, A.T., Vitousek, P., 1998. Nutrient dynamics on a precipitation gradient in Hawai'i.Oecologia 113, 519–529.

Bai, Y., Zhang,W., Jia, X., Wang, N., Zhou, L., Xu, S., Wang, G., 2010. Variation in root: shootratios induced the differences between above and belowground mass–density rela-tionships along an aridity gradient. Acta Oecol. 36, 393–395.

Bonan, G.B., 1995. Land-atmosphere CO2 exchange simulated by a land surface processmodel coupled to an atmospheric general circulation model. J. Geophys. Res. 100,2817–2831.

Calzadilla, A., Rehdanz, K., Betts, R., Falloon, P., Wiltshire, A., Tol, R.S.J., 2013. Climatechange impacts on global agriculture. Clim. Chang. 120, 357–374.

Cao, J.-J., Xiong, Y.-C., Sun, J., Xiong, W.-F., Du, G.-Z., 2011. Differential benefits of multi-and single-household grasslandmanagement patterns in the Qinghai-Tibetan Plateauof China. Hum. Ecol. 39, 217–227.

Chen, J., Liu, J., Cihlar, J., Goulden, M., 1999. Daily canopy photosynthesis model throughtemporal and spatial scaling for remote sensing applications. Ecol. Model. 124,99–119.

Chen, B., Chen, J.M., Ju, W., 2007. Remote sensing-based ecosystem–atmosphere simula-tion scheme (EASS)—Model formulation and test with multiple-year data. Ecol.Model. 209, 277–300.

Chen, S., Chen, J., Lin, G., Zhang, W., Miao, H., Wei, L., Huang, J., Han, X., 2009. Energy bal-ance and partition in Inner Mongolia steppe ecosystems with different land usetypes. Agric. For. Meteorol. 149, 1800–1809.

Duan, Q., Xin, X., Yang, G., Chen, B., Zhang, H., Yan, Y., Wang, X., Zhang, B., Li, G., 2011. Cur-rent situation and prospect of grassland management decision support systems inChina. Computer and computing technologies in agriculture IV. Springer, New York,NY, USA, pp. 134–146.

Falkowski, P., Scholes, R., Boyle, E.A., Canadell, J., Canfield, D., Elser, J., Gruber, N., Hibbard,K., Högberg, P., Linder, S., 2000. The global carbon cycle: a test of our knowledge ofearth as a system. Science 290, 291–296.

Fan, L.-Y., Gao, Y.-Z., Brück, H., Bernhofer, C., 2009. Investigating the relationship betweenNDVI and LAI in semi-arid grassland in Inner Mongolia using in-situ measurements.Theor. Appl. Climatol. 95, 151–156.

Fang, J., Piao, S., Field, C.B., Pan, Y., Guo, Q., Zhou, L., Peng, C., Tao, S., 2003. Increasing netprimary production in China from 1982 to 1999. Front. Ecol. Environ. 1, 293–297.

Fang, J., Guo, Z., Piao, S., Chen, A., 2007. Terrestrial vegetation carbon sinks in China,1981–2000. Sci. China Ser. D Earth Sci. 50, 1341–1350.

Feng, X., Liu, G., Chen, J., Chen, M., Liu, J., Ju, W., Sun, R., Zhou, W., 2007. Net primary pro-ductivity of China's terrestrial ecosystems from a process model driven by remotesensing. Journal of Environmental Management 85, 563–573.

Feng, Z., Yang, Y., Zhang, Y., Zhang, P., Li, Y., 2005. Grain-for-green policy and its impactson grain supply in West China. Land Use Policy 22, 301–312.

Friedlingstein, P., Joel, G., Field, C., Fung, I., 1999. Toward an allocation scheme for globalterrestrial carbon models. Glob. Chang. Biol. 5, 755–770.

Gao, Z., Liu, J., 2008. Simulation study of China’s net primary production. Chinese ScienceBulletin 53, 434–443.

Garbulsky, M.F., Paruelo, J.M., 2004. Remote sensing of protected areas to derive baselinevegetation functioning characteristics. J. Veg. Sci. 15, 711–720.

Gosling, S.N., Arnell, N.W., 2013. A global assessment of the impact of climate change onwater scarcity. Clim. Chang. 1–15.

Hargreaves, G., Allen, R., 2003. History and evaluation of Hargreaves evapotranspirationequation. J. Irrig. Drain. Eng. 129, 53–63.

Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G., Jarvis, A., 2005. Very high resolution in-terpolated climate surfaces for global land areas. Int. J. Climatol. 25, 1965–1978.

Janssens, I.A., Pilegaard, K., 2003. Large seasonal changes in Q10 of soil respiration in abeech forest. Glob. Chang. Biol. 9, 911–918.

Jentsch, A., Beierkuhnlein, C., 2008. Research frontiers in climate change: effects of ex-treme meteorological events on ecosystems. Compt. Rendus Geosci. 340, 621–628.

Ji, J., Huang, M., Li, K., 2008. Prediction of carbon exchanges between China terrestrial eco-system and atmosphere in 21st century. Sci. China Ser. D Earth Sci. 51, 885–898.

Ju, W., Chen, J.M., 2005. Distribution of soil carbon stocks in Canada's forests andwetlandssimulated based on drainage class, topography and remotely sensed vegetation pa-rameters. Hydrol. Process. 19, 77–94.

Ju, W., Chen, J.M., Black, T.A., Barr, A.G., Liu, J., Chen, B., 2006. Modelling multi-yearcoupled carbon and water fluxes in a boreal aspen forest. Agric. For. Meteorol. 140,136–151.

Ju, W., Chen, J.M., Black, T.A., Barr, A.G., McCaughey, H., 2010. Spatially simulating changesof soil water content and their effects on carbon sequestration in Canada's forests andwetlands. Tellus B 62, 140–159.

Knapp, A.K., Smith, M.D., 2001. Variation among biomes in temporal dynamics of above-ground primary production. Science 291, 481–484.

Knapp, A.K., Beier, C., Briske, D.D., Classen, A.T., Luo, Y., Reichstein, M., Smith, M.D., Smith,S.D., Bell, J.E., Fay, P.A., 2008. Consequences of more extreme precipitation regimes forterrestrial ecosystems. Bioscience 58, 811–821.

Li, F.R., Kang, L.F., Zhang, H., Zhao, L.Y., Shirato, Y., Taniyama, I., 2005. Changes in intensityof wind erosion at different stages of degradation development in grasslands of InnerMongolia, China. J. Arid Environ. 62, 567–585.

Li, L., Vuichard, N., Viovy, N., Ciais, P., Ceschia, E., Jans, W., Wattenbach, M., Béziat, P.,Gruenwald, T., Lehuger, S., 2011. Importance of crop varieties andmanagement prac-tices: evaluation of a process-based model for simulating CO2 and H2O fluxes at fiveEuropean maize (Zea mays L.) sites. Biogeosciences Discussions 8, 2913–2955.

Li, X.R., Jia, X.H., Dong, G.R., 2006. Influence of desertification on vegetation pattern vari-ations in the cold semi-arid grasslands of Qinghai-Tibet Plateau, Northwest China.J. Arid Environ. 64, 505–522.

Liu, J., Diamond, J., 2005. China's environment in a globalizing world. Nature 435,1179–1186.

Liu, J., Chen, J.M., Cihlar, J., 2003. Mapping evapotranspiration based on remote sensing:an application to Canada's landmass. Water Resour. Res. 39.

Liu, M., Jiang, G., Li, L., Li, Y., Gao, L., Niu, S., 2004. Control of sandstorms in Inner Mongolia,China. Environ. Conserv. 31, 269–273.

Liu, J., Zhang, Z., Xu, X., Kuang, W., Zhou, W., Zhang, S., Li, R., Yan, C., Yu, D., Wu, S., 2010.Spatial patterns and driving forces of land use change in China during the early 21stcentury. J. Geogr. Sci. 20, 483–494.

Liu, X., Duan, L., Mo, J., Du, E., Shen, J., Lu, X., Zhang, Y., Zhou, X., He, C., Zhang, F., 2011. Ni-trogen deposition and its ecological impact in China: an overview. Environ. Pollut.159, 2251–2264.

Liu, Y., Liu, R., Chen, J.M., 2012. Retrospective retrieval of long-term consistent global leafarea index (1981–2011) from combined AVHRR and MODIS data. J. Geophys. Res.Biogeosci. 117, G04003.

Liu, Y., Ju, W., He, H., Wang, S., Sun, R., Zhang, Y., 2013. Changes of net primary productiv-ity in China during recent 11 years detected using an ecological model driven byMODIS data. Front. Earth Sci. 7, 112–127.

Lu, Y., Zhuang, Q., Zhou, G., Sirin, A., Melillo, J., Kicklighter, D., 2009. Possible decline of the car-bon sink in the Mongolian Plateau during the 21st century. Environ. Res. Lett. 4, 045023.

Luo, Y., Gerten, D., Le Maire, G., Parton, W.J., Weng, E., Zhou, X., Keough, C., Beier, C., Ciais,P., Cramer, W., 2008. Modeled interactive effects of precipitation, temperature, and[CO2] on ecosystem carbon and water dynamics in different climatic zones. Glob.Chang. Biol. 14, 1986–1999.

Matsushita, B., Tamura, M., 2002. Integrating remotely sensed data with an ecosystemmodel to estimate net primary productivity in East Asia. Remote Sens. Environ. 81,58–66.

McMichael, A.J., Powles, J.W., Butler, C.D., Uauy, R., 2007. Food, livestock production, ener-gy, climate change, and health. Lancet 370, 1253–1263.

Melillo, J.M., Steudler, P.A., Aber, J.D., Newkirk, K., Lux, H., Bowles, F.P., Catricala, C., Magill,A., Ahrens, T., Morrisseau, S., 2002. Soil warming and carbon-cycle feedbacks to theclimate system. Science 298, 2173–2176.

Mitchell, S.W., Csillag, F., 2001. Assessing the stability and uncertainty of predicted vege-tation growth under climatic variability: northern mixed grass prairie. Ecol. Model.139, 101–121.

Mu, Q., Zhao, M., Running, S.W., Liu, M., Tian, H., 2008. Contribution of increasing CO2 andclimate change to the carbon cycle in China's ecosystems. J. Geophys. Res. Biogeosci.2005–2012, 113.

Mu, S.J., Zhou, S.X., Chen, Y.Z., Li, J.L., Ju, W.M., Odeh, I.O.A., 2013. Assessing the impact ofrestoration-induced land conversion and management alternatives on net primaryproductivity in Inner Mongolian grassland, China. Glob. Planet. Chang. 108, 29–41.

Ni, J., 2000. Net primary production, carbon storage and climate change in Chinese bi-omes. Nord. J. Bot. 20, 415–426.

Ni, J., 2002. Carbon storage in grasslands of China. J. Arid Environ. 50, 205–218.Ni, J., 2003. Plant functional types and climate along a precipitation gradient in temperate

grasslands, northeast China and southeast Mongolia. J. Arid Environ. 53, 501–516.

93Y. Chen et al. / Rangeland Ecology & Management 69 (2016) 84–94

Ni, J., 2004a. Estimating net primary productivity of grasslands from field biomass mea-surements in temperate northern China. Plant Ecol. 174, 217–234.

Ni, J., 2004b. Forage yield-based carbon storage in grasslands of China. Clim. Chang. 67,237–246.

Parmesan, C., 2007. Influences of species, latitudes and methodologies on estimates ofphenological response to global warming. Glob. Chang. Biol. 13, 1860–1872.

Pärtel, M., Helm, A., Reitalu, T., Liira, J., Zobel, M., 2007. Grassland diversity related to theLate Iron Age human population density. J. Ecol. 95, 574–582.

Pei, S., Fu, H., Wan, C., 2008. Changes in soil properties and vegetation following exclosureand grazing in degraded Alxa desert steppe of Inner Mongolia, China. Agric. Ecosyst.Environ. 124, 33–39.

Peng, S., Piao, S., Wang, T., Sun, J., Shen, Z., 2009. Temperature sensitivity of soil respirationin different ecosystems in China. Soil Biol. Biochem. 41, 1008–1014.

Piao, S., Fang, J., Zhou, L., Guo, Q., Henderson, M., Ji, W., Li, Y., Tao, S., 2003. Interannual var-iations of monthly and seasonal normalized difference vegetation index (NDVI) inChina from 1982 to 1999. J. Geophys. Res. Atmos. 1984–2012, 108.

Piao, S., Fang, J., Zhou, L., Ciais, P., Zhu, B., 2006a. Variations in satellite-derived phenologyin China's temperate vegetation. Glob. Chang. Biol. 12, 672–685.

Piao, S., Friedlingstein, P., Ciais, P., Zhou, L., Chen, A., 2006b. Effect of climate and CO2

changes on the greening of the Northern Hemisphere over the past two decades.Geophys. Res. Lett. 33.

Piao, S., Mohammat, A., Fang, J., Cai, Q., Feng, J., 2006c. NDVI-based increase in growth oftemperate grasslands and its responses to climate changes in China. Glob. Environ.Chang. 16, 340–348.

Piao, S., Fang, J., Zhou, L., Tan, K., Tao, S., 2007. Changes in biomass carbon stocks in China'sgrasslands between 1982 and 1999. Glob. Biogeochem. Cycles 21.

Piao, S., Ito, A., Li, S., Huang, Y., Ciais, P., Wang, X., Peng, S., Nan, H., Zhao, C., Ahlström, A.,2012a. The carbon budget of terrestrial ecosystems in East Asia over the last two de-cades. Biogeosciences 9, 3571–3586.

Piao, S., Tan, K., Nan, H., Ciais, P., Fang, J., Wang, T., Vuichard, N., Zhu, B., 2012b. Impacts ofclimate and CO2 changes on the vegetation growth and carbon balance of Qinghai–Tibetan grasslands over the past five decades. Glob. Planet. Chang. 98–99, 73–80.

Porporato, A., D’odorico, P., Laio, F., Rodriguez-Iturbe, I., 2003. Hydrologic controls on soilcarbon and nitrogen cycles. I. Modeling scheme. Adv. Water Resour. 26, 45–58.

Potter, C.S., Randerson, J.T., Field, C.B., Matson, P.A., Vitousek, P.M., Mooney, H.A., Klooster,S.A., 1993. Terrestrial ecosystemproduction: a processmodel based on global satelliteand surface data. Glob. Biogeochem. Cycles 7, 811–841.

Schapendonk, A.H.C.M., Stol, W., van Kraalingen, D.W.G., Bouman, B.A.M., 1998. LINGRA, asink/source model to simulate grassland productivity in Europe. European Journal ofAgronomy 9, 87–100.

Schimmelpfennig, S., Müller, C., Grünhage, L., Koch, C., Kammann, C., 2014. Biochar,hydrochar and uncarbonized feedstock application to permanent grassland—Effectson greenhouse gas emissions and plant growth. Agric. Ecosyst. Environ. 191, 39–52.

Scurlock, J., Hall, D., 1998. The global carbon sink: a grassland perspective. Glob. Chang.Biol. 4, 229–233.

Shen, M., Tang, Y., Chen, J., Zhu, X., Zheng, Y., 2011. Influences of temperature and precip-itation before the growing season on spring phenology in grasslands of the centraland eastern Qinghai-Tibetan Plateau. Agric. For. Meteorol. 151, 1711–1722.

Sitch, S., Friedlingstein, P., Gruber, N., Jones, S.D., Murray-Tortarolo, G., Ahlström, A., Doney, S.C.,Graven, H., Heinze, C., Huntingford, C., 2013. Trends and drivers of regional sources andsinks of carbon dioxide over the past two decades. Biogeosci. Discuss. 10, 20113–20177.

Stephens, S.L., Burrows, N., Buyantuyev, A., Gray, R.W., Keane, R.E., Kubian, R., Liu, S.R.,Seijo, F., Shu, L.F., Tolhurst, K.G., van Wagtendonk, J.W., 2014. Temperate and borealforest mega-fires: characteristics and challenges. Front. Ecol. Environ. 12, 115–122.

Sui, X., Zhou, G., 2013. Carbon dynamics of temperate grassland ecosystems in China from1951 to 2007: an analysis with a process-based biogeochemistry model. Environ.Earth Sci. 68, 521–533.

Sui, X., Zhou, G., Zhuang, Q., 2013. Sensitivity of carbon budget to historical climate vari-ability and atmospheric CO2 concentration in temperate grassland ecosystems inChina. Clim. Chang. 117, 259–272.

Wang, X., Lu, C., Fang, J., Shen, Y., 2007. Implications for development of grain-for-greenpolicy based on cropland suitability evaluation in desertification-affected northChina. Land Use Policy 24, 417–424.

Wang, S., Wilkes, A., Zhang, Z., Chang, X., Lang, R., Wang, Y., Niu, H., 2011. Managementand land use change effects on soil carbon in northern China's grasslands: a synthesis.Agric. Ecosyst. Environ. 142, 329–340.

Wang, L., Zhou, Y., Zhou,W., Wang, S., 2013. Fire danger assessment with remote sensing:a case study in Northern China. Nat. Hazards 65, 819–834.

Wang, T., Lin, X., Peng, S., Cong, N., Piao, S., 2014. Multimodel projections and uncer-tainties of net ecosystem production in China over the twenty-first century. Chin.Sci. Bull. 59, 4681–4691.

Warren, C., Dreyer, E., Adams, M., 2003. Photosynthesis-Rubisco relationships in foliage ofPinus sylvestris in response to nitrogen supply and the proposed role of Rubisco andamino acids as nitrogen stores. Trees 17, 359–366.

Xia, J., Luo, Y., Wang, Y.P., Weng, E., Hararuk, O., 2012. A semi-analytical solution to accel-erate spin-up of a coupled carbon and nitrogen land model to steady state. Geosci.Model Dev. Discuss. 5, 803–836.

Xing, X., Xu, X., Zhang, X., Zhou, C., Song, M., Shao, B., Ouyang, H., 2010. Simulating net pri-mary production of grasslands in northeastern Asia using MODIS data from 2000 to2005. J. Geogr. Sci. 20, 193–204.

Yang, K., Koike, T., Ishikawa, H., Kim, J., Li, X., Liu, H., Liu, S., Ma, Y., Wang, J., 2008a. Turbu-lent flux transfer over bare-soil surfaces: characteristics and parameterization. J. Appl.Meteorol. Climatol. 47, 276–290.

Yang, Y., Fang, J., Tang, Y., Ji, C., Zheng, C., He, J., Zhu, B., 2008b. Storage, patterns and con-trols of soil organic carbon in the Tibetan grasslands. Glob. Chang. Biol. 14,1592–1599.

Yang, Y., Fang, J., Ma, W., Guo, D., Mohammat, A., 2010. Large-scale pattern of biomasspartitioning across China's grasslands. Glob. Ecol. Biogeogr. 19, 268–277.

Zhang, Y., Zhou, G., 2012. Primary simulation on the response of leaf maximumcarboxylation rate to multiple environmental factors (in Chinese). Chin. Sci. Bull.57, 11–12.

Zhang, Q., S.W., Zhao, D., Dai, E., et al., 2013a. Temporal-spatial Changes in Inner Mongo-lian Grassland Degradation during Past Three Decades. Agricultural Science & Tech-nology 14, 7.

Zhang, N., Zhao, Y.-S., Yu, G.-R., 2009. Simulated annual carbon fluxes of grassland ecosys-tems in extremely arid conditions. Ecological Research 24, 185–206.

Zhang, F., Ju, W., Shen, S., Wang, S., Yu, G., Han, S., 2013b. How recent climate change in-fluences water use efficiency in East Asia. Theor. Appl. Climatol. 116, 359–370.

Zheng, Z.-M., Yu, G.-R., Fu, Y.-L., Wang, Y.-S., Sun, X.-M., Wang, Y.-H., 2009. Temperaturesensitivity of soil respiration is affected by prevailing climatic conditions and soil or-ganic carbon content: a trans-China based case study. Soil Biol. Biochem. 41,1531–1540.

94 Y. Chen et al. / Rangeland Ecology & Management 69 (2016) 84–94